Impacts of climatic and non-climatic risks are on the rise in the major maize-growing counties of western Kenya. We conducted a questionnaire survey of sampled maize farmers in Nandi, West Pokot, Uasin Gishuand Trans Nzoia counties, achieving 210 responses. We used a stepwise regression model to evaluate the factors influencing farmers' perceptions of farming risks. Results show that most of the sampled farmers were aware of the risks, and perceived reduced rainfall with erratic patterns to be the major climatic risk in crop production. The non-climatic factors were identified as inadequate farm size, limited extension services, land degradation and low soil fertility. The determinant factors that influence farmers' perception of climatic and non-climatic risks affecting crop production were age, farm size, income, crop production cost, marital status, the highest level of education and farming experience. Understanding the risks faced in crop production and determinants of farmers' perceptions can be important in the development and dissemination of sustainable agronomic strategies tailored towards improving crop production.

Globally and over a longer period, climate change continues to pose negative impacts to the agricultural sector. In particular, Sub-Saharan Africa (SSA) has been well known as a continent that is highly susceptible to climate change due to over reliance on rainfed agriculture, multiple stressors and low capacity to cope. Land use/cover change is also a locally pervasive and worldwide trend that has notable implications on ecological trends, sustainability of ecosystem services, communities and biodiversity. Kenya, like other countries in SSA, has experienced remarkable and far-reaching land use/cover changes over time as a consequence of variable climate, population pressure, environmental degradation, land fragmentation and unsustainable agricultural practices that have contributed to food insecurity and increased vulnerabilities. However, information is scant on how climate variability and land use/cover changes could affect current and future ecological suitability for production of main food crops under rainfed conditions. Thus, to bridge the gap, this study analysed the trend of climate variability and land use/cover changes and their impacts using a case study of western Kenya, an area of high agricultural potential, as a classic example of an affected region.

The outcome of a systematic review of 186 journal articles published in different parts of the world show a projected reduction in maize yields of between 8-38% by the end of the 21^st century due to climate change. In Africa, an ecological niche modelling using the Maximum Entropy (MaxEnt) approach to examine land suitability for major food crops identified two major shifts: 11.1-22.0% expansion in areas suitable for production of maize, and a 1.6-7.3% decline in areas suitable for production of millet and sorghum. In Kenya, the assessment showed a potential for increase in unsuitable areas for maize production by an average of between 1.9–3.9% and a decrease of moderately suitable areas by 14.6–17.5%. The change in the suitable and highly suitable areas in the country is an increase of between 17–20% and 9.6%, respectively, under climate change. The loss of suitability for production of food crops is likely to cause detrimental impacts on food security for the communities as uncertainties of projected climate variability and change unfold

Hydrological modeling of the temporal trends of rainfall variability using Mann-Kendall test and Sen's slope estimator in the major maize growing counties of western Kenya revealed noticeable decrease of March-April-May rainfall, and an increase in September-October-November rainfall. Under RCP 8.5, the results show a projected seasonal shift and an increase in intensity of major rains from March-April-May to June-July-August, which is likely to affect climatic suitability for cultivation and production of key crops in the region. The multi-index characterization of drought on a 12-month time series show a likelihood of moderate to extreme drought years under RCP 4.5 and RCP 8.5 climatic scenarios.

The western Kenya region has also experienced spatial and temporal land use/cover changes of varying trends and magnitudes in the years 1995, 2001, 2010 and 2017. Classification of multi-spectral Landsat images show that between 1995 and 2001, there was an increase in builtup areas by 71%, forestland by 43%, farms by 5%" and decrease in grassland by 47%. By 2017, the built-up areas had increased by 225% and farms by 17%" while the forestland, grassland and water reduced by 38%, 10% and 11%, respectively. The observed changes were characterised by increased settlements and encroachment of sensitive ecosystems.

The use of GIS-based Revised Universal Soil Loss Equation (RUSLE) modelling approach and remote sensing techniques to examine the impacts of land use/cover changes on land degradation due to sheet, rill and inter-rill soil erosion processes in western Kenya resulted to average soil loss of 0.3 ton/ha/year and 0.5 ton/ha/year, in the years 1995 and 2017, respectively. Of the total soil loss, farms contributed more than 50%, both in 1995 and 2017 followed by grass/shrub (7.9% in 1995 and 11.9% in 2017), forest (16% in 1995 and 11.4% in 2017), and the least in built up areas. The highest soil erosion rates were observed in farms cleared from forests (0.84 tons/ha) followed by those converted from grass/shrub areas (0.52tons/ha). The rate of soil erosion was observed to increase with slope due to high velocity and erosivity of the run-off. Soils susceptible to highest erosion rates are found primarily in slopes of more than 30 degrees, especially in Mt. Elgon, Chereng'anyi hills and Elgeyo escarpments.

The maize farmers in western Kenya perceived reduced rainfall with erratic patterns to be the major climatic risk affecting crop production. The non-climatic factors were identified as inadequate farm size, limited extension services, land degradation and low soil fertility. The major adaptation strategies undertaken by the farmers included change in planting dates by either planting early or late during a season, diversification of crops, growing early maturing cultivars and use of drought-tolerant varieties. The use of logistic and multiple linear regression models revealed the key determinants of adaptation strategies by the farmers to include farm size, income and extension training. Understanding farmers' responses to climate change in rainfed crop production systems could assist in planning adaptation strategies towards sustainable crop production.

The findings of this thesis are subject to uncertainties, which are associated with the modelling tools used, reliability and quality of climatic and crop occurrence data. Future changes in climatic scenarios could result to changes in bioclimatic variables, causing different shifts in climatic suitability and would be a consideration in future investigation. In addition, the use of one region as a case study for the scope on local perceptions limit the diversity of the results. Thus, future work can explore additional diversified cases to create a collection for comparison across regions, ecological and climatic zones.

Overall, this research provides knowledge and information on how climate variability and land use/cover changes affect rainfed agriculture. Such knowledge provides a wider perspective of the issues alongside the local perceptions that are inherent in addressing the associated challenges and in decision making related to land use planning, land degradation management, drought preparedness and adaptation of crop production under climate change.